Purchase this article with an account.
Takuma Morimoto, Kazuho Fukuda, Keiji Uchikawa; Luminance-balance Based Estimation of an Illuminant in Chromatically Biased Scenes. Journal of Vision 2016;16(4):31. doi: 10.1167/16.4.28.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
When an illuminant changes both chromaticity and luminance of surfaces change. However, despite a substantial amount of color constancy algorithm, most methods rely on chromaticity change of a surface to estimate the illuminant. Based on the optimal color hypothesis, Uchikawa et al. (2012) recently revealed the importance of luminance-balance of surfaces although luminance-balance had smaller effect on estimating an illuminant than that of chromatic change. Since red or blue surfaces become lighter when an illuminant changes to red or blue, respectively, we assumed the larger number of reddish and bluish surfaces in a scene might enhance the effect of luminance-balance. In order to address this question we employed (1) balanced, (2) red-blue dominated and (3) yellow-green dominated sets of surrounding stimuli. In order to simulate the illuminant changes with no chromatic change of surrounding surfaces we manipulated only luminance-balance of each surface while its chromaticity was kept constant. Observers adjusted both chromaticity and luminance of the center test stimulus so that it looked as a full-white paper under the test illuminant. It was again shown that luminance-balance is useful to estimate the illuminant, but illuminants were estimated as being shifted towards the mean chromaticity of the scene in the red-blue and the yellow-green dominated scenes. Importantly, results showed better color constancy for the red-blue than the yellow-green dominated scene, indicating that the effect of luminance-balance depended on the chromaticity available in the scene. This suggests that the optimal color hypothesis would account for the mechanism to estimate an illuminant.
This PDF is available to Subscribers Only